Purchase this article with an account.
Valerie Goffaux, Bruno Rossion; Faces are “spatial”- Holistic perception of faces is subtended by low spatial frequencies. Journal of Vision 2005;5(8):540. doi: 10.1167/5.8.540.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Faces are processed as wholes more than as collections of elements. The holistic perception of a face is so robust that it influences the processing of its features. Both the whole-part advantage (Tanaka & Farah, 1993) and the composite-face effect (Young et al., 1987) illustrate this point. In the whole-part paradigm, subjects have to recognize (or match) facial features (eyes, nose, and mouth) presented either in a face, or in isolation. Holistic processing is indexed by the better processing of features when they are embedded in a facial context, than when they are presented in isolation. In the composite-face paradigm, the top half of a face is joined with the bottom half of another face, creating a composite face. Observers have difficulty finding that two top halves are identical if the bottom halves belong to different faces, supporting the view that the composite face is perceived holistically, i.e. as a new whole face.
Here, we used these 2 paradigms with spatially filtered faces to test the hypothesis that face holistic processing is supported by information conveyed in low spatial frequencies of the stimulus (Sergent, 1986; Morrison & Schyns, 2001). The whole-part and composite-face effects were measured with faces filtered to preserve the low spatial frequencies (32 cycles/image HSF), or the full spectrum of image luminance variations.
Both the whole-part and the composite-face effects were significantly larger with LSF as compared to HSF faces. While the whole-part effect observed for LSF faces had a similar magnitude as for full spectrum faces, the composite-face effect was larger for LSF faces than for full spectrum faces.
These results suggest that the holistic processing of faces is mostly subtended by coarse information, as provided by LSF. They also yield perspectives about how the various face cues (holistic, featural, etc) might integrate over time to build a robust face representation.
This PDF is available to Subscribers Only